Metabolic diseases present particular difficulty for clinicians because they are often present for years before becoming clinically apparent. Clinical predictors of future diabetes mellitus (DM) are imperfect. The identification of individuals at ris is of particular importance because the delay or prevention of type 2 DM is possible via both behavioral and pharmacological interventions. In the first 3 1/2 years of this grant, we applied our mass spectrometry-based metabolomics platform to identify and validate metabolite profiles of those destined to develop overt DM. The findings were based on longitudinal follow-up in two prospective cohorts, the Framingham Heart Study and the Malmo Diet and Cancer Study. The strongest individual predictors of future DM included branched chain amino acids, aromatic amino acids, and 2-aminoadipic acid (a lysine degradation product). These metabolites were elevated up to 12 years before the onset of DM in glucose-tolerant individuals at baseline, and predicted DM above and beyond clinical risk factors and biochemical markers. These findings suggest that metabolite profiling could play a role in DM prevention strategies. To test this hypothesis, we will leverage the unique resources of the Diabetes Prevention Program (DPP), a multi-center RCT of interventions to prevent or delay the onset of DM in individuals at high-risk of the disease. In Aim 1, we will assess the relation of baseline metabolites with incident DM, and test interactions of metabolites with therapy. In Aim 2, we will determine whether lifestyle and pharmacologic interventions lead to changes in selected metabolites. In Aim 3, we will investigate relations of metabolites with insulin sensitivity and insulin secretion measures. We will perform cross-sectional and longitudinal analyses to investigate the relation of metabolites with measures of insulin sensitivity and secretion. The DPP represents an ideal setting in which to conduct the investigations for the next phase of our grant, examining novel predictors of DM and their potential role in personalizing therapeutic interventions. PUBLIC HEALTH RELEVANCE: Current treatments for diabetes mellitus (DM) are only partially successful, in part because they are based on limited knowledge of its root causes. Furthermore, there is no way to accurately predict who will develop DM, thus limiting our ability to intervene effectively. Our goal is to use new blood chemical profiling approaches to illuminate our understanding of the underlying disease mechanisms in DM, to identify novel predictors of DM, and to evaluate whether the predictors can help to personalize therapeutic interventions.